Working Parser?

[Motenten]

1.3.0 (March 16, 2009)
* Major refactoring of core elements for piping messages through the program.
* Rewrite of Offense and Defense tab accumulators. Old accumulators had some issues, new ones should be significantly faster.
- Fix issue with charmed players not being reported properly. General bug fix on resolving entity types.
- General bug fixing of architecture issues.


I'm pretty sure this will resolve pretty much all CPU-time-eating issues for the Offense and Defense tabs, reducing time consumed by 1-2 orders or magnitude most of the time. Other tabs may be updated if this goes well. If you still have issues with KParser bogging your system down while having any/all of Experience/Offense/Defense tabs open (no guarantees on some of the more esoteric ones), please let me know.

Available at: kparser - Google Code

Edit: Bug fixed, re-released 1.3.0a.

Further edit: Version 1.3.0c will break some things regarding timestamps (they're now stored in UTC instead of local time) in order to pave the way to certain planned improvements. 1.3.0a is still available on the downloads page, but isn't featured, should you wish to use that instead.

[Motenten]

1.3.1 (March 16, 2009)
+ Add function to reparse older parses and update their timestamp information.
+ Fixed reparsing function to carry over user-edited player information.
- Using 'Save As' and trying to overwrite an existing file works now.

Available at kparser - Google Code

[Motenten]

One more update today.

1.3.2
Fixed a bug that would cause long reparse runs to be terminated prematurely.
Improved coding in Thief plugin SATA detection. Should also be much faster when displaying single mobs (such as during an active parse).

[Motenten]

1.3.2.1 -- Fix a major bottleneck in message insertion into the database. This speeds up reparsing tremendously, and may also ameliorate (maybe fix? I can only hope) the problem of the program slowing down the system after very long sessions.

[Raineer]

I know vaguely where it could happen, but haven't figured out how it can happen, and haven't replicated it yet. Will be keeping an eye out to fixing it, though.

Hobbyist programmer here, let me know if there's a way I can help debug this with you. Shows up roughly 50% of the times I start the program. If it occurs and I leave it open for more than, say, 3 minutes after a Zerg (sometimes during the attempt) the program will lock up with no error. It just locks like a normal frozen Windows program. If I kill the process and re-open the parse data it is fully valid and complete (despite being unable to switch to it during the kill or right after.)

I need to download the latest version you posted, though. Need to make sure it's still occurring.

[Motenten]

Version 1.3.3 release (@ http://code.google.com/p/kparser/ ). Adds an interesting new feature -- custom mob filtering.

Most plugins that make use of a mob filter (except DPM) have a new option giving access to a dialog box that lets you hand-pick exactly which mobs/fights you want to include for that plugin's analysis.

The selected mobs are kept globally; that is, changes you make when viewing one tab affect all tabs that are currently set to use the custom filter.

There are a few buttons on the dialog box for simple selections, and there's a context menu in the listbox itself that lets you make affect other subsets of mobs. Those include: check/uncheck all of the currently selected mob type (you must have clicked on a mob in the list box for it to count as 'selected'); check/uncheck all of the selected mob type that have the same base XP; and check/uncheck all mobs below the current selection (eg: skip past the first few mobs you don't want, select everything below that point, then at the bottom unselect any mobs that were fought after the end of the segment you want analyzed).

Use it, abuse it, and let me know if you encounter any problems. Also let me know if there are any other custom context menu selection sets that you want.


@Raineer: yes, if that's still occurring with the latest builds, let me know. It would be rather difficult for me to tell you how to debug the issue, though; you'd want to install VS2008, download the source, and do a debug trace on what was being processed, which probably will be rather difficult to follow. If you're willing to look into it, though, I'd appreciate any help you can give in tracking this one down.

[Motenten]

Version 1.3.4: Fix bug in combobox updating in the DPM plugin.

[Kohan]

In case this is useful to anyone -- more likely the end users rather than the programmer, however -- I'd like to jot down my own adventures with this program, and getting it to function correctly. I'd downloaded the most recent version, and attempted to run it on Windows XP x64.

Originally, I'd installed SQLCE 3.5 SP1 (both x86 and x64), but K-Parser would only return errors at that point. None of the tabs appeared, and a new parsing log couldn't be started.

At this point, I uninstalled the x86 version of SQLCE 3.5 SP1, and downloaded the prior version mentioned in the readme, plain ol' 3.5. I installed that.

In addition, I -- having already installed .NET 3.5 previously -- downloaded and installed the NDP35SP1-KB958484-x64 update, which can be found here:

Microsoft .NET 3.5 Family Update

It now works just fine for me.

[Raineer]

@Raineer: yes, if that's still occurring with the latest builds, let me know. It would be rather difficult for me to tell you how to debug the issue, though; you'd want to install VS2008, download the source, and do a debug trace on what was being processed, which probably will be rather difficult to follow. If you're willing to look into it, though, I'd appreciate any help you can give in tracking this one down.

Haven't seen it again but I've played on a very limited basis recently. I have VS2008 on another PC, I'll put it on my FFXI one and do a general debug if I see it again.

Thanks for all the hard work!

[Motenten]

Version 1.3.5

- Fixes the annoying problem of losing preferences when updating the parser. Old settings are now carried over when using a new version of the program.
- Fixes an issue with multiple redraws when Group flag is changed.
* Custom mob filter shows fight number to match the dropdown listing.
- Dancer Steps and Flourishes are properly categorized as Enfeebles rather than ordinary damaging abilities.

[Epical]

Thanks for your continued work on Kparser, Motenten. Even though there may be no replies, everyone is thankful for the updates. =)

[Araelus]

Motenten, thanks for all the work you've done. I have one question - is there a way to fix the errors on "Edit Mob Filter" and "Group Mobs" by xp return?

It lists the true xp return in the Fights tab, but in "Group Mobs" and "Edit Mob Filter" the (xp) value is incorrect. - More specifically, some 72xp mobs are being listed as 70xp mobs, and some 72xp mobs are being listed as 65xp mobs. No idea why.

Thanks.

[Motenten]

1.3.6 released. Available at kparser - Google Code

Changes:

1.3.6 (March 30, 2009)
- Bug fix: Failed enhancements (eg: blindna failed to take effect) were incorrectly parsed and ended up causing a false CharmedPlayer to show up in the logs, and thus in the fight lists. Reparsing is necessary to get corrected data.
- Bug fix: Crash bug due to empty battle data when changing selected mob.
* Fights tab now shows fights where you received XP, but were not in range to see what was killed.
* Fights tab shows the chain number for each battle.
- Fix issues in creating new battle entries in situations where the mob was killed but not named (ie: outside of range of seeing what was killed).
- Fix issue of corrected base XP value (after removing chain multiplier) being applied to fights of chain 0. The full spread of proper xp value should now show up.

This should fix the issue you're seeing, Araelus.

[Araelus]

Youuuuu are the best! Thank you!

This update fixed the error where Seigan Counters were being listed twice, too. :D

The pDIF testing guys are in your debt~

[Hogie]

Im sure i missed it in this thread somewhere but i couldnt find it. I didnt see anything in read-me of kparser either but is there a way to select instance? Or can you only run this with 1 instance open?

[Motenten]

In the Options dialog (Tools/Options...) when set up for reading from RAM, check the box for 'Specify process when starting'. That will allow you to specify which instance of FFXI it tries to read when parsing.

[Araelus]

Quick question, Motenten: I noticed the "Offense Detail" tab produces different results between the "Show Detail" block of numbers and the automatically generated frequency chart.

Out of 9885 attack rounds, all of them show up in the "Show Detail" number block, eg.

      69   76   87   87   93   75   84   75   64   80
      72   80   84   92   73   80   60   62   68   61
      90   93   69   80   63   66   86   87   76   80
      54   69   65   95   55   58   85   93   89   71
      92   61   71   89   78   75   70   68   67   78
      87   95   63   87   68   77   54   78   72   57
      86   82   86   63   77   61   55   74   92   53

when I use Excel to produce a histogram, but if I copy the K-Parser bin/freq, eg.

105: 2
106: 4
107: 8

values and sum the frequencies, I get only 4181 attack rounds.

What am I doing wrong?

[Motenten]

Sorry Araelus, I'm not seeing that. Sample parse, 1 person with 179 melee attacks (16 crit) shows 163 melee (not-crit) attacks in the detail block, and the frequency sum matches that. Checked several others and getting the same results.

If possible, can you attach a copy of the parse to an issue report on the google page? Save without chat if you like, and zip it if it's large.

[Araelus]

I clicked "reparse" a few times and it sorted itself out. :/ Glad to see it isn't broken, but dunno why the problem occurred at all. Thanks. ^^;

[shaddix]

Nice to see kparser have such great dev support.